List your company

Open-ended generation on SciQ

5.21ECEGrACE
3.176416.903230.6344.3568Sep 11, 2025

Evaluation results

MethodLinks
GrACEBackbone=Phi3-3.8B2025.095.2116.9382.3758.1
GrACEBackbone=Llama2-7B2025.095.4514.3387.7257.4
ApricotBackbone=Phi3-3.8B2025.097.0918.876.45—
P(True)Backbone=Llama2-7B2025.097.4624.9449.41—
GrACEBackbone=Llama3.1-8B2025.098.1619.1375.6263.5
ApricotBackbone=Llama2-7B2025.098.4216.6982.04—
ActCabBackbone=Phi3-3.8B2025.098.7222.270.24—
Platt ScalingBackbone=Llama3.1-8B2025.099.0323.3769.46—
ActCabBackbone=Llama2-7B2025.099.3424.5160—
Platt ScalingBackbone=Llama2-7B2025.0910.1224.8461.41—
Platt ScalingBackbone=Phi3-3.8B2025.0910.8324.5773.69—
ActCabBackbone=Llama3.1-8B2025.0910.9921.6569.12—
Seq. LikelihoodBackbone=Phi3-3.8B2025.0914.4723.8173.6958.5
Seq. LikelihoodBackbone=Llama2-7B2025.0916.5226.7461.4157.2
ApricotBackbone=Llama3.1-8B2025.0917.0624.1769.94—
P(True)Backbone=Llama3.1-8B2025.0917.9627.952.83—
VerbalBackbone=Llama3.1-8B2025.0921.1126.264.09—
Seq. LikelihoodBackbone=Llama3.1-8B2025.0924.6327.2669.4662.8
P(True)Backbone=Phi3-3.8B2025.0931.8230.0979.66—
VerbalBackbone=Phi3-3.8B2025.0936.5236.5556.44—
VerbalBackbone=Llama2-7B2025.0956.0537.9453.07—